A proposition of interior air flow assessment method for humid tropical architecture.

Increasing demands for energy saving and a higher degree of comfort in rooms compels designers or architects to use more sophisticated analysis methods. The measurement in situ, numerical simulation (CFO), and wind tunnel investigations are three of methods which are always utilised to analyse or to assess air flow in rooms and their environment. However, these methods remain generally very difficult for the majority of the designers or the architects. With the help of these two latter methods we attempt to establish a database concerning the influence of architectural design elements on interior air flow; with which we afford the training phase of our artificial neural nets. This paper presents an assessment method of interior air flow using artificial neural nets. The air flow distribution inside a building depends not only on the external wind velocity, but also largely on the building design overall especially for humid tropical architecture. Due the difficulty to evaluate the interior air flow if we have to take into account a number of architectural design elements, we proposed to use this approach. The utilisation of the neural networks as a universal predictor is an interesting field of investigation. They provide reliable results in the cases where many parameters have to be taken into account simultaneously.